They also work on universal scales again for the most part.
The universe having more mass or more energy makes more sense than random gravity, or a phenomenon that is completely not-understood on the same scale as space-time curvature, the latter being mostly because we should observe it also locally.
Non-Constant G just breaks the universe, it's also not exactly "Non-Constant" it's still constant just at a different value as far as dark energy goes.
But considering that we calculate the value correctly on a local scale it makes very little sense that is changes all of a sudden especially considering that it changes the same everywhere including within our galaxy.
Dark matter is there to fix another very different picture because without it you'll have an almost completely random gravitational constant even on small scales - star systems, and galaxies, this simply can't be the case.
Even without going into relativity if you have a classical system (stars) that doesn't account for the motion you are missing mass some where.
Now there are other attempts to explain some of these effects e.g. the gravitational waves from super massive black holes that are slamming into each other, new experiments predominately those which will measure pulsars are will attempt to discover and measure those with much greater accuracy which might account for some of the randomness but won't account for the fact that we need more mass and energy in the universe to make things work.
Science has always used placeholders and mechanisms for things that should be there but weren't discovered, this is the great thing of good science because it's predictive, and more often than not it works, even when you don't understand the entire picture.
A great example of this is the periodic table it was constructed by pure observation of the attributes of various elements, it predated the standard model, the discovery of protons, neutrons, isotopes and the such. However it was so correct that it didn't matter because the principle worked and the extrapolated attributes of the elements were enough to build a system that didn't had to be changed when you figured out how all the internal gadgets work.
Now the Precession of Mercury is actually a great example, Newtonian laws of motion account for just under 95% of Mercury's orbital velocity, relativity (the fact that mass curves space) accounts for the missing 5%.
Before we understood relativity we had planet Vulcan which was the doodad that science constructed to explain the orbital mechanics of Mercury.
The big difference between Mercury and Dark Matter/Energy is that we need to find something that works on all scales, the need for dark matter and dark energy comes from relative small scales as well as very big ones, now there could be some other force or phenomenon we completely don't understand but you also can't look for something completely unknown. You would be hard pressed to look at something like the orbit of mercury and come up with relativity as an explanation that just wont work, you'll never end up with that as a result. Dark Matter or Dark Energy for the other hand are much wider place holders than a Planet X.
On top of that Dark Matter is not only needed to explain relativity but is also predicted by other theories mainly big bang nucleosynthesis, it may also be needed (or could account for) various observations of gravitational microlensing since MACHO's don't seem to be able to account for all of those.
And as far as WIMP's go well they aren't hacks, these are predicted by quite a few of nucleosynthesis theories that go beyond the standard model, we already have weak interactive particles, but they all tend to be extremely light however mass doesn't seem to play a role in their nature so WIMP's could very well exists, people didn't think neutrino's were real either.